[Source: PRNewswire-FirstCall/] – SGIC high-performance compute (HPC) systems have long been a fixture in universities throughout the world, speeding time to discovery across a wide range of disciplines.
SGI shared-memory servers can be found in university research facilities throughout the world, but a new trend is prompting universities to migrate from smaller systems that serve individual departments to larger clusters and shared-memory systems capable of meeting the needs of many departments and disciplines. This trend is saving IT administrators the cost and time of managing multiple small systems, and it offers a new generation of researchers the kind of HPC resources they need to tackle ever-larger bioinformatics problems.
Universities are choosing SGI systems to meet the demanding challenges of bioinformatics for several reasons.
The SGI server architecture offers the ability to scale compute power by adding processors as needed, without a forklift upgrade.
SGI systems meet the need for both shared memory and cluster computing environments to maximize application performance and enhance workflows.
Both shared memory and cluster capabilities can be combined to create a seamless, hybrid environment.
“Over the last several years, a number of universities have begun purchasing larger SGI Altix multiprocessor systems and have added IT departments to manage their use within the central data center,” said Deepak Thakkar, Ph.D. higher education and research solutions manager, SGI. “Many researchers and departments then use grant monies and university matching funds to purchase processors on these large systems. Their purchased processors are always there for their needs, and occasionally, depending on the volume from other departments, IT managers can make more processors available to specific projects.”
Recent examples of this trend include Michigan State University, Universite de Montreal, and the University of Arizona, all which are using SGI Altix systems for cutting edge bioinformatics research.
Michigan State University — Gene Mutation Research
Michigan State University (MSU) has used an SGI(R) Altix(R) XE1300 cluster with 1,024 cores since September 2007. Evolutionary biologist Barry Williams, assistant professor, Zoology and Microbiology and Molecular Genomics, purchased 16 cores of the SGI Altix XE system to identify which forces of nature, such as natural selection or changes in population size, cause changes in DNA to accumulate over time, as well as those forces that prevent new mutations that arise in natural populations. When Williams sees mutations that seem to have some effect on genotype, he makes those mutations in many different strains of yeasts, and puts them in different environments to verify their potential effects. Williams works primarily with easily manipulated single-cell yeast microbes.
While he has done work on the university’s existing SGI(R) Altix(R) 3700 HPC system at MSU, installed some two years ago, Williams considers the number of projects he can run in parallel on the new SGI Altix XE cluster as a boon to his research.
“It’s not unusual for one analysis running on one processor to take several days to a week. We have 6,000 genes that make up the yeast genome that we want to analyze, and each one can take a day. Without the Altix cluster this just becomes impossible,” said Williams, who also conducts artificial life research on the SGI system. “The artificial life evolves autonomous computer programs in silico — we do this to make general predictions about the evolutionary process that we can next test in the yeasts. The beauty of artificial life research is that you can repeat the experiment of evolution with the exact same starting conditions and let evolution take place again and again. I have 16 processors on the systems, so at any one time I can run 16 completely different worlds in parallel. They all start at the same point; they all have the same genetic diversity; they all start with the same individuals and the same conditions, but simultaneously, 16 times, I can let them evolve for millions upon millions of generations.”
Many other MSU departments including Nuclear Physics, Mathematics and Chemistry will also be harnessing the power of the SGI system (see separate release, April 29, 2008).
Universite de Montreal — World’s Largest Heart Simulation Model with 2 Billion Elements
In the quest to discover how the mechanisms of heart disease work, researchers at the Universite de Montreal (UdM) ran the largest mathematical simulation of a heart ever assembled — a 2 billion element model — on a high-performance SGI(R) Altix(R) 4700 system with 1.2TB of memory. (See separate release January 15, 2008) The SGI Altix is believed to be the largest shared memory computing system in Canada
Until recently, the largest heart models in the world had at most a few million elements.
Dr. Mark Potse and Dr Alain Vinet, both affiliated with the Research Center of Sacre-Coeur Hospital and the Biomedical Engineering department at UdM, began running 100 to 120 million-point models as part of their heart disease research. Potse and Vinet regularly use 60 of the 768 Intel(R) Itanium(R) 2 processors running on the SGI Altix which, as part of the Quebec Network for High-Performance Computing (RQCHP), is shared by many researchers from across Canada since it’s unveiling in February 2007.
In October last year, Potse and Vinet had the opportunity to run their custom electrocardiography (ECG) code using the entire SGI Altix system with all 786 processors and all 1.2TB of shared memory. Originally written by them on an earlier SGI system and ported to the SGI Altix system’s Linux(R) environment in 2003, the ECG code made the leap from 120 million points to 2 billion with ease.
“This was a test to see if the simulation works and to determine that, if we have a much bigger machine, our software will be able to run more efficiently,” said Potse. “This capability is really for the future when we can use this size of machine on a regular basis, but with the Altix system we have made the heart model of the future today.”
The new UdM model is up to 1,000 times more detailed than previous models, enabling new scientific discoveries that would never be possible via observation alone.
University of Arizona — Maize Sequencing Research
The University Information Technology Services (UITS) at the University of Arizona (UA)) purchased two SGI Altix 4700 systems with a total of 1.2TB of memory over a year ago. Instead of departments purchasing processors before installation, the UA allows professors and researchers to add processors to the scalable SGI compute environment. The systems are a university-wide resource available to any department in need of HPC, which includes a number of ongoing and new bioinformatics projects.
Research Professor Cari Soderlund heads the Arizona Genomics Computational Laboratory (AGCoL) that is currently working on the computational aspects of sequencing 30,000 genes for maize. In order to determine the sequence of genes, subsequences of about 800 base pairs (bp) are generated. To reconstruct the original gene sequence, the sub sequences are analyzed to determine the overlapping sub-sequences. For the 700,000 sequences, the global shared memory of the SGI Altix system will accelerate data generation and analysis.
“As our lab is often processing large datasets, the speedup from the SGI Altix system should increase the turn-around time between the generation of biological data and computational analysis,” said Soderlund, who recently began porting her project to the SGI Altix system. “It would also allow for more experimentation of parameters and algorithms for larger datasets.”
“The Altix 4700 has been our workhorse for the past year and is used by researchers throughout campus, including life sciences,” said Dr. Michael Bruck, Assistant Director of Research Computing at the UA’s UITS, which coordinates central research computer resources on campus. “The Altix is meeting and exceeding our expectations.”
SGI Altix systems empower research at many prestigious universities, including UC Riverside, Purdue University, Stony Brook University, University of Utah, Technische Universitat Dresden (TU Dresden), University of Sao Paulo and U.S. Air Force Academy.
“With many universities deploying both shared memory systems and clusters such as the SGI Altix servers and Altix XE clusters, we are now seeing jobs scheduled based upon the architecture that best suits their computing needs-an entirely seamless process for the researchers,” added Thakkar. “SGI systems are being rapidly adopted not only for bioinformatics, but for all disciplines ranging from astronomy to zoology.”